Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore the foundational principles of generative models in this 45-minute conference talk delivered at the Computational Genomics Summer Institute 2025. Delve into the core concepts that underpin generative modeling approaches, examining how these mathematical frameworks can be applied to create and understand complex data distributions. Learn about the theoretical underpinnings that drive modern generative techniques, including diffusion models and their applications in various domains. Discover insights into preference learning methodologies and how win rate concepts can simplify complex learning paradigms. Examine advanced topics such as inference-time scaling and steering mechanisms for diffusion models, understanding how these techniques can be optimized for better performance. Investigate automated learning approaches for multivariate diffusions, including strategies for determining optimal diffusion locations, methods, and recovery processes. Gain valuable perspectives on the mathematical foundations that enable generative models to capture and reproduce intricate patterns in high-dimensional data spaces, with particular relevance to computational genomics applications.
